Improving facilitated modelling

By Vincent de Gooyert

Vincent de Gooyert (biography)

Here I explore two outcomes of facilitated modelling – cognitive change and consensus forming – and ask: how can achieving those outcomes be improved?

But first, what is facilitated modelling?

Facilitated modelling is an approach where operational researchers act as facilitators to model an issue collaboratively with stakeholders, usually in a workshop. Operational research, also known as operations research, seeks to improve decision-making by developing and applying analytical methods.

Two central aims of facilitated modelling are to achieve cognitive change and to form consensus.

Cognitive change is the idea that participants of facilitated modelling workshops come in with a certain worldview, and that the intervention leads them to learn about the issue and accordingly change their minds. The intention of facilitated modelling is that the participants change their minds in such a manner that, after the workshop, their view on the problem is more similar to those of the other participants than before the workshop. In other words: facilitated modelling helps to form consensus.

What can go wrong in facilitated modelling?

Each participant in a facilitated modelling workshop brings his or her mental model about the issue under consideration. This mental model consists of all the deeply held beliefs that individuals maintain. Engaging in a structured dialogue can lead to instances where opposing mental models are confronted with each other. Cognitive change requires participants to surface and resolve these differences leading to an improved understanding of the issue.

If, however, participants avoid discussing sensitive issues, differences in mental models may remain hidden reducing the cognitive change achieved.

Consensus forming can be interactional or mental. Interactional consensus occurs as part of the workshop discussion process and may be more or less explicit. For example, the chair may claim consensus has been achieved and then move on to another topic. Alternatively, the chair may provide an opportunity to express dissent and make sure that all participants have a chance to articulate their views.

Mental consensus refers to congruence or alignment of beliefs among workshop participants. While interactional consensus is observed by meeting participants, mental consensus is observed by an outsider (eg., a researcher) by aggregating individual measures.

There can be a mismatch between mental consensus and perceived interactional consensus, which is referred to as:

  • pluralistic ignorance if the majority agrees but thinks they do not
  • false consensus if participants think they agree but in fact they do not.

Our research (de Gooyert et al., 2022) has shown that there is little empirical evidence about the occurrence of cognitive change and consensus formation. Further, in a study gathering such evidence ourselves, we found that although cognitive change and consensus forming were achieved, experienced (perceived interactional) and observed (mental) cognitive change were not related.

Improving facilitated modelling

Central to improving facilitated modelling is to gather more empirical data about the cognitive change and consensus forming outcomes, especially under different workshop conditions. The aim is to determine what makes workshops effective in achieving these outcomes.

Does it help, for example:

  • if the facilitated modelling workshops are preceded by a round of interviews where sensitive issues are specifically explored to ensure that they are also included in the workshop?
  • if workshops avoid having dual aims of team building and an analytical consideration of the issue of interest? In team building there is a tendency to avoid uncomfortable discussions which stops sensitive issues from being raised.
  • if workshop participants are informed about the results of research examining outcomes? Workshop participants may suffer from an illusion of productivity, to explain why participants are convinced they are more productive in a brainstorming group than when working individually and pooling their results. Being informed about what actually happened may make workshop participants more discerning about the workshops they participate in and how they contribute.

What do you think? Do you have other evidence or ideas to contribute? How do these issues play out in the participatory modelling community?

To find out more:

de Gooyert, V., Rouwette, E., van Kranenburg, H., Freeman, E. and van Breen, H. (2022). Cognitive change and consensus forming in facilitated modelling: A comparison of experienced and observed outcomes. European Journal of Operational Research, 299, 2: 589-599. (Online – open access) (DOI):
This paper also provides the references for the ideas cited in this i2Insights contribution.

Biography: Vincent de Gooyert PhD is an associate professor at Nijmegen School of Management, Radboud University in Nijmegen, the Netherlands. His research and teaching revolve around increasing understanding of, as well as intervening in, societal transformations towards sustainability, using and contributing to methods on stakeholder engagement, system dynamics, and socio-technical transitions.

10 thoughts on “Improving facilitated modelling”

  1. Thanks Vincent. Your first proposal, and to a certain extent also the second one, refer to the need to stimulate and manage productive cognitive conflict within the group. Research shows that moderate levels of cognitive conflict can be beneficial to group performance (as long as interpersonal conflict remains low), and facilitated modelling (FM) can certainly support this. All FM approaches (whether decision-centered as in Decision Conferencing, or system-centered as in Group Model Building) seek to reveal, contrast and integrate conflicting views, beliefs, priorities, mental models, etc. Whether the model-based activity in and by itself can do this is, however, open to question. Our research on self-facilitated groups using models to support their decisions shows that a large proportion of groups never surface cognitive conflict. Consequently, the role of the facilitator in both stimulating cognitive conflict and fostering productive conflict management behaviors becomes critical. Your first proposal provides one way of doing this prior to an FM session. But you could also support conflict management during the session deliberately and systematically.

    Your last proposal raises another important issue. Trying to measure cognitive change, learning,etc through self reported questionnaires is always tricky and not that reliable. Studies of transfer of learning using discrete event simulation models suggest an alternative approach that could be developed for the FM case. Our research has shifted the focus to measuring interactional impacts observed in situ, as they are demonstrable not only to the researchers, but most importantly to the participants themselves. And these impacts have real consequences for action.

    In relation to the above comments, you may find of interest the following sources:
    Franco, L. Alberto, Engin, Aysegul, Ayşegül & Rouwette, Etiënne (2023). TRacing conflcit management paths in a model-supproted environment. In Yu, Maemura, Yu, Horita, Masahide, Fang, Liping & Zaraté, Pascale (Eds) Proceedings of the 23rd International Conference on Group Decision and Negotiation, Tokyo, 11-15 June.

    Franco, L. Alberto, & Greiffenhagen, Christian. (2018). Making OR practice visible: Using ethnomethodology to analyse facilitated modelling workshops. European Journal of Operational Research, 265(2), 673-684. doi: 10.1016/j.ejor.2017.08.016

    Franco, L. Alberto, & Nielsen, Mie Femø. (2018). Examining group facilitation in situ: The use of formulations in facilitation practice. Group Decision and Negotiation, 27(5), 735-756. doi: 10.1007/s10726-018-9577-7

    Franco, L. Alberto, Rouwette, Etiënne AJA, & Korzilius, Hubert. (2016). Different paths to consensus? The impact of need for closure on model-supported group conflict management European Journal of Operational Research, 249(3), 878-889. doi: 10.1016/j.ejor.2015.06.056

    Monks, Thomas, Robinson, Stewart, & Kotiadis, Kathy. (2016). Can involving clients in simulation studies help them solve their future problems? A transfer of learning experiment. European Journal of Operational Research, 249(3), 919-930. doi:

    Tavella, Elena, & Franco, L Alberto. (2015). Dynamics of group knowledge production in facilitated modelling workshops: An exploratory study. Group Decision and Negotiation, 24(3), 451-475. doi:

  2. Very nice, but I still can’t quite understand why do we need so many terms for pretty much the same approach. We used to have companion modeling, group model building, then came shared vision planning and mediated modeling, then collaborative modeling. At one point participatory modeling was proposed as an overarching name that does not belong to any particular group or organization. Now we also have facilitated modeling, which again seems to be talking more or less about the same. What’s the advantage? How do our users decide which clone of participatory modeling they want to implement? Is this useful to have so many different names for pretty much the same technique?

    At first I thought that facilitated modeling is the participatory modeling that is focused on cognitive change and changing minds. But then it appears that it’s also about “communication, learning, consensus, behavioural change and implementation”. But that seems to cover pretty much the whole domain. Then what’s special?

    • Thank you Alexey, and yes, I agree, there are many commonalities and the differences in terminology can be very confusing. It would be very helpful if there would be just one shared language that everyone would adhere to. At the same time, I’m also recognizing that there is a lot of diversity in this field, with different communities working along similar lines, all with their own rich backgrounds and experiences that lead to terms having different connotations that are not shared by everyone. It would also be a pity if the richness of that diversity is lost through imposing terminology standards, perhaps it could lead to potentially fruitful avenues being overlooked.

      • Yes, agreed. Looks like this could be a good topic for an overview paper. Who is who and what is what? How do we navigate through the diversity of terms and connotations? Anybody interested in joining? Would be good to attract representatives of all the different communities and make some kind of an overview and perhaps chart a dichotomy or a decision tree to help people figure out where do they want to belong. Drop me an e-mail if interested.

  3. I wonder whether an LLM (large language model), like Chat GPT could assess the transcripts of such group deliberations to yield a more insightful assessment of consensus or understanding. Perhaps providing a channel for 1:1 side conversations/inquiries between participants and the AI (artificial intelligence) could yield greater insights.

    • Yes it would be very interesting to tap into the potential of AI along these lines. Interesting idea of having AI being one of the participants of a facilitated modeling session! That way, it could help feed the discussion, but (other) experts would still be able to guide the conversations/inquiries.

  4. Thanks Vincent for the useful blog. I wonder if consensus is only one of the outcomes that FM aim to. For example, a workshop could help participants appreciate where others are coming and develop empathy of their concerns without aiming for consensus.

    • Good to hear you find it useful. Indeed, there are many outcomes that facilitated modeling can achieve, and often times multiple goals are combined (either explicitly or implicitly). I agree that developing empathy can be an important goal as well. In our paper we mention a few, namely communication, learning, consensus, behavioural change and implementation, and we focus on cognitive change and consensus forming also because of practical reasons: because these are two goals that we could measure both in terms of observed and experienced results.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: